SnipFind Any Word in Any Recording
Upload a recording — or a whole folder of voice messages — type a word or phrase, and jump straight to every moment it was said. No scrubbing through an hour of audio. 100% in your browser.
Have voice messages from someone you've lost? Drop in all their WhatsApp or voicemail recordings at once. Search for "I love you," "I'm proud of you," or your own name — and hear every time they said it. Or browse the word cloud of everything they ever told you, and click a word to play those moments back. Nothing is uploaded — their voice stays on your device.
Drop your recordings here
or click to browse. Add one file or many — they're searched together. Up to ~60 minutes total.
No file handy? — a short clip with some swearing to test the censor.
100% in your browser. Your audio stays on your device. The Whisper AI model downloads once (~75 MB) from a public CDN, then runs locally. We never see your recordings. Privacy policy.
We transcribe your audio once (free, in your browser), then you can search it as many times as you like. Keep this tab open while it runs — bigger files take longer. The model is cached after the first download.
What was said most
Click words to add them to your searchSearch inside your audio — find every moment a word was spoken
The Audio Word Finder transcribes your recording with OpenAI's open-source Whisper model running entirely in your browser, then builds a searchable, time-stamped index of every word. Type what you're looking for and it lists each moment it was said — with a play button that jumps straight to that exact spot, plus a couple of seconds of run-up so you hear it in context.
One recording or a hundred
Drop in a single hour-long interview, or a whole folder of short voice notes. Multiple files are joined into one timeline and transcribed together, and every result is labelled with which file it came from and the timestamp inside it — so "Voicemail 4 · 0:38" tells you exactly where to look.
Who it's for
- Journalists & researchers — find every time a name, claim, or keyword comes up across hours of interviews.
- Students & note-takers — jump to where the lecturer explained a specific term.
- Podcasters & editors — locate a quote or a stumble to cut, without scrubbing the timeline.
- Families — gather a loved one's voice messages and revisit the things they said most, or search for a phrase that meant something to you.
Completely private
Because the AI runs on your own device, nothing is ever uploaded. That matters for confidential interviews and therapy notes — and it matters even more for irreplaceable personal recordings. The model file downloads once from a public CDN; after that everything works offline.
How accurate is it?
Very good for clear speech. Accuracy drops with heavy background noise, music, overlapping speakers, or strong accents. If a word was mumbled it may be transcribed slightly differently — try a close spelling, or browse the word cloud to see how a word actually came through.